Test the idea, not the ad ID.
Most creative testing ends with a ranked list of ad IDs and no lesson that carries into the next flight. Assets here are registered in the creative graph with concept, message, offer, format and visual-treatment tags, so results roll up to the part that is reusable. A test compares tagged concepts against one audience definition under one spend rule, and grades them on attributed outcomes and CRM quality instead of platform clicks. Variants showing fatigue are retired inside the window, so decay is not scored as a losing concept.
Tags attached before the first impression
A variant is registered with its concept, message, offer and format before it ever reaches a network, which rules out tagging a result after you have seen it. The finding then reads "the guarantee-led concept wins on square video" instead of naming four ad IDs that will not exist next quarter.
Same audience, same spend rule
Concepts under test share an audience definition and a fixed spend rule for the full window, so no variant wins by being handed more delivery. Google, Meta and TikTok keep running auction-level delivery and bidding inside each cell; the test controls only what enters the comparison and for how long.
Graded on customers, not clicks
Outcomes are read through the attribution service and your CRM, so a concept whose clicks are cheap and whose leads stall is separated from one credited with customers who stay. Refunds and margin come from your finance system and are applied to the same cohort before any concept is called a winner.
In this capability
- Concept-level rollups
- Message and offer tags
- Format and placement splits
- Downstream quality readout
Experiments
The rest of Experiments
Audience Tests
Expansion, suppression and seed tests, graded on which segments ended up as paying customers.
Channel Tests
Budget allocation and market-entry tests: what a channel is worth at a different spend level, or at all.
Landing-Page Tests
Message and page variants joined to outcomes through content IDs and your web analytics.
Incrementality
Geo and holdout designs run where volume supports them, reported separately from attribution credit.
Learning Agenda
A ranked queue of open business questions, so testing is a programme rather than an occasional habit.
One goal. Every channel. Real business truth.
Connect your ad accounts, say what a good customer is worth, and let the Growth Brain work the objective inside limits you set.
No ad spend commitment · Connect one channel to start · [email protected]